Social Cost of Carbon (SCC)

• Social Cost of Carbon (SCC) is the monetized estimate of the marginal global economic damages from an additional tonne of CO₂ emissions, calculated using integrated assessment models.
• SCC estimation integrates climate science, welfare economics, and risk analysis to link emissions, temperature changes, and resulting economic impacts.
• Practical applications include informing optimal carbon taxation, regulatory evaluations, liability frameworks, and urban mitigation strategies amid deep uncertainties.

The social cost of carbon (SCC) is the present value of the marginal global economic damages resulting from the emission of one additional tonne of carbon dioxide or its equivalent at a specific point in time. SCC operationalizes the externalities of carbon emissions in monetary terms and serves as a central benchmark for climate policy, including optimal carbon taxation, regulatory evaluation, intergovernmental transfers, and liability calculations. SCC estimation integrates welfare economics, climate science, macroeconomics, and risk analysis via integrated assessment models (IAMs), but also reflects deep uncertainty, ethical debates about time preference, and the practical realities of implementing mitigation and adaptation policies.

1. Formal Definition and Theoretical Basis

The SCC can be formally expressed as the marginal welfare loss, in consumption-equivalent units, due to a one-tonne emission at date $t$. In representative-agent IAMs (e.g., DICE, FUND), this is written as

$$\mathrm{SCC}_t = -\frac{\partial W}{\partial E(t)} / U'(C(t))$$

where $W$ is the social welfare function, $E(t)$ is emissions at time $t$, and $U'(C(t))$ is the marginal utility of consumption. In dynamic programming terms, the SCC is the shadow price of the atmospheric carbon stock, normalized to the numeraire (usually consumption or capital): $$\mathrm{SCC}_t = -1000 \cdot \frac{\partial V_t/\partial M_{\mathrm{AT}, t}}{\partial V_t/\partial K_t}$$ with $V_t$ the value function in the state vector $(K, \mathbf{M}, \mathbf{T}, \ldots)$ (Cai et al., 2015, Tol, 2023, Khabarov et al., 2020).

SCC aggregates discounted future climate damages over the time path induced by a marginal emission at time zero: $$\mathrm{SCC} = \int_0^\infty \frac{\partial C(t)}{\partial E(0)} e^{-\rho t} dt$$ Here, $$\mathrm{SCC}_t = -\frac{\partial W}{\partial E(t)} / U'(C(t))$$0 is the social discount rate, embodying both pure rate of time preference (PRTP, $$\mathrm{SCC}_t = -\frac{\partial W}{\partial E(t)} / U'(C(t))$$1) and the elasticity of intertemporal substitution ($$\mathrm{SCC}_t = -\frac{\partial W}{\partial E(t)} / U'(C(t))$$2), typically via the Ramsey formula $$\mathrm{SCC}_t = -\frac{\partial W}{\partial E(t)} / U'(C(t))$$3, with $$\mathrm{SCC}_t = -\frac{\partial W}{\partial E(t)} / U'(C(t))$$4 the consumption growth rate (Tol, 2024).

2. Core Modeling Frameworks and SCC Calculation

Quantification of SCC entails linking emissions to climate dynamics and economic damages, and then discounting the resulting loss trajectory:

• Emission to climate system: SCC models utilize carbon-cycle modules (e.g., three-box Maier-Reimer–Hasselmann) to translate emissions into atmospheric concentrations, and simple or complex energy-balance models to link concentrations to temperature anomalies (Cai et al., 2015, Smith et al., 2023).
• Climate to damages: Damages are typically functions of temperature anomaly (e.g., quadratic in temperature, as $$\mathrm{SCC}_t = -\frac{\partial W}{\partial E(t)} / U'(C(t))$$5) with empirically calibrated coefficients (Tol, 2023, Dong et al., 2024). Emerging metanalytic frameworks adopt Bayesian-averaged or meta-regressed damage functions (Agrawala et al., 20 Jan 2026).
• Discounting damages: Future damages are discounted using $$\mathrm{SCC}_t = -\frac{\partial W}{\partial E(t)} / U'(C(t))$$6, fundamentally altering SCC estimates. Representative-agent models use a single exponential discount rate, while new approaches implement declining (hyperbolic) discounting deriving from heterogeneous or uncertain rates (Dong et al., 3 Feb 2025).
• Value extraction: In practice, SCC is calculated as the perturbed net present value of welfare (or output), often via a finite-difference approach: simulate the model baseline, then simulate again with an emissions pulse and compute the difference in discounted output streams (Murakami et al., 16 Apr 2025, Tol, 2023, Khabarov et al., 2020).

Integrated Assessment Models (IAMs) such as DICE, DSICE, FUND, and PAGE provide the dominant computational infrastructure, with increasing attention to incorporating stochastic economic growth, climate tipping elements, income heterogeneity, and regionalization (Cai et al., 2015, Cai et al., 2023, Agrawala et al., 20 Jan 2026, Estrada et al., 2024).

3. Uncertainty, Risk, and Distributional Properties

Uncertainty is multidimensional in SCC estimation, encompassing:

• Climate and carbon-cycle uncertainty: Variability in equilibrium climate sensitivity (ECS), carbon uptake rates, and forcing translates into substantial SCC range. For state-of-the-art ensembles using, e.g., FaIR v2.1 with calibrated ECS and aerosol ERF, the 5–95% SCC interval in stringent mitigation scenarios can span an order of magnitude (e.g., USD 821–4,434/tCO₂ in a 1.5°C target scenario) (Smith et al., 2023, Folini et al., 2021).
• Damage-function uncertainty: The high-damage tail is particularly sensitive to structural choices (quadratic, convex, fat-tailed, or growth-channel damages), as shown by Bayesian Model Averaging of candidate functions (Agrawala et al., 20 Jan 2026, Tol, 2024).
• Discounting: Both the mean and the shape (tail thickness) of the SCC distribution are highly sensitive to discount-rate assumptions. Lower PRTP and lower inverse EIS result in higher and more right-skewed SCC (Tol, 2 Jul 2025, Tol, 2024).
• Economic risk, risk preferences, and tipping points: Models incorporating recursive utility (Epstein–Zin), stochastic growth, and Markovian, irreversible climate tipping elements produce median and upper-tail SCCs that are multiples of deterministic benchmarks; e.g., including stochastic tipping elevates 2005 SCC from USD 38/tC to USD 189/tC, with the upper 1% at USD 2,500/tC by 2100 (Cai et al., 2015).
• Heterogeneity and the “Weitzman premium”: Population, income, and time-preference heterogeneity lead to hyperbolic discounting, amplifying SCC by factors of 5–200 relative to average-agent models. In long horizons, the smallest discount rate dominates, and the “most patient” sub-population shapes the aggregate SCC (Dong et al., 3 Feb 2025).

Meta-analyses confirm that the SCC distribution is highly right-skewed, with modes near USD 80/tCO₂, means exceeding USD 200/tC (USD 56/tCO₂) and extreme values into the thousands (Tol, 2024, Tol, 2 Jul 2025, Tol, 2021). Distributional properties are further impacted by scenario structure, regional aggregation, and publication bias.

4. Key Empirical Estimates and Recent Trends

Meta-analytical techniques, statistical emulators, and scenario-based modeling collectively provide the foundation for empirical SCC estimation:

• Central tendency and dispersion: Recent meta-analyses yield a mean SCC of approximately USD 207/tC, with a modal value of USD 25–50/tC and a large tail above USD 500/tC (Tol, 2024). The median SCC in reweighted meta-emulation, aligning discounting and risk aversion assumptions with expert evidence, is USD 229/tC (USD 62/tCO₂), with a 90% confidence band of ±10/tC (Tol, 2 Jul 2025).
• Time evolution: Controlling for discount rate, mean SCC has increased fourfold over the past decade (e.g., from USD 33/tC to USD 146/tC at PRTP = 3%; and from USD 446/tC to USD 1925/tC at PRTP = 0%) (Tol, 2021).
• Scenario dependence: Under optimal pathways from process-based IAMs (e.g., under the “net-zero by 2050” and “net-zero by 2100” scenarios), SCC increases from USD 30–50/tCO₂ in 2025 to USD 140–430/tCO₂ by 2100, with the highest values under fossil-driven SSPs (Murakami et al., 16 Apr 2025).
• Urban, regional, and phase-space estimates: Urbanized areas (cities) account for ≳90% of SCC under spatially explicit IAMs, with per-tonne values from USD 137/tCO₂ (regional warming only) to USD 1,075/tCO₂ with urban heat island and persistence (Estrada et al., 2024). Scenario-based frameworks like OPTiMEM frame SCC not as a point but as a multi-dimensional economic phase space, conditional on scenario, long-run real rate, and time window (Hanley et al., 3 Feb 2026).

A summary table of typical SCC ranges under representative assumptions: | Framework/scenario | Central SCC (USD/tCO₂ or tC) | Range | |-------------------------------|------------------------------|-----------------------------------| | Meta-analysis (mean) | USD 207/tC | Mode: 25–50/tC; tail >500/tC | | DICE-SSP, 2025–2100 | 30–50 → 140–430/tCO₂ | Across SSPs/IAM calibrations | | DSICE (risk), 2005–2100 | 38–189 (“baseline”); | 1% tail up to 2,500 by 2100 | | Urban IAM (CLIMRISK) | 137–1,075/tCO₂ | ~90% urban | | OPTiMEM (phase-space) | 3D surface | Scenarios/discount d/n variation |

5. Normative, Methodological, and Policy Debates

The SCC is both conceptually and methodologically contested:

• Marginal vs. comprehensive cost: SCC is fundamentally a marginal-damage price, not the total cost of mitigation and adaptation. Uniform carbon pricing covering total abatement and legacy damages (the “polluter pays” principle) is an order of magnitude higher (e.g., implied total-cost CO₂ price USD 500/tCO₂ vs. SCC ≈ USD 50/tCO₂ in DICE-2016) (Fries, 2023).
• Discounting and its ethical basis: The tension between prescriptive and descriptive time preference, and debates over market-based versus social discount rates, induces persistent spread in SCC estimates. Approaches tying discount rates to market yields (e.g., via very long-term “carbon bonds”) or population preference heterogeneity offer alternatives but generate additional uncertainty and complexity (Dong et al., 3 Feb 2025, Hanley et al., 3 Feb 2026).
• Climate module calibration and model misspecification: Miscalibrated climate blocks (e.g., in DICE-2016) can lead to SCC values with extreme comparative statics; robust calibration to CMIP5/CMIP6 benchmarks is essential (Folini et al., 2021).
• Winsorizing and economic feasibility constraints: Conceptual winsorizing removes implausibly high outliers (e.g., SCC > USD 2×10⁸/tC), imposing hard caps by maximum ability to pay (GDP/emissions) or tax-revenue feasibility. Even after winsorizing, mean SCCs remain robust (e.g., USD 375/tC Weitzman-bound, USD 221/tC Hobbes-bound) and central policy proposals (USD 50–200/tCO₂) are not affected (Tol, 10 Aug 2025).
• Retrospective updating and market solutions: ReSCCU and retroactive pricing schemes propose hedgeable, market-linked updates to SCC as empirical damages are observed, managed through insurance-market platforms and prediction markets (Bengio et al., 2022).
• Regional, income-weighted, and liability-based SCCs: Aggregate global SCC disguises cross-country heterogeneity—global North (OECD) values can be ~4–5 times higher than the South, with striking implications for cross-border transfers and fair carbon pricing (Dong et al., 2024). Liability frameworks that allocate “net liability” based on the harm done by a country’s emissions to others minus the harm suffered from others’ emissions reveal that middle-income, carbon-intensive economies are net payers, while poorest and richest countries are net recipients (Agrawala et al., 20 Jan 2026).

6. Applications and Policy Implementation

SCC directly informs:

• Optimal carbon taxation and regulatory cost-benefit analysis (e.g., via the US interagency SCC).
• Design of cap-and-trade and hybrid carbon pricing: In partial/non-cooperation, optimal regional carbon taxes are generally set as the difference between marginal abatement cost and permit price (Cai et al., 2023).
• Liability and compensation regimes: Net liabilities derived from national SCCs can guide international transfer schemes or “loss and damage” negotiations, reflecting both self-harm and externalities (Agrawala et al., 20 Jan 2026).
• Urban adaptation policy: Quantification of urban SCC and the social cost of urban heat island motivates targeted adaptation and mitigation investments, with net present benefits per urban dweller of USD 484–1,562 for a 1% permanent UHI reduction (Estrada et al., 2024).
• Scenario-based and phase-space analysis: Economic, physical, and discounting uncertainties are increasingly acknowledged via scenario surfaces or full distributional approaches (meta-emulation, phase-space, insurance pricing), rather than by point estimates (Tol, 2 Jul 2025, Hanley et al., 3 Feb 2026).

7. Research Frontiers and Open Issues

Key ongoing and emerging directions include:

• Integrating climate–economy feedbacks: Enhanced explicit modeling of feedbacks (e.g., energy return on energy invested, climate shocks on growth) and non-linear damages (Hanley et al., 3 Feb 2026, Smith et al., 2023).
• Refined uncertainty quantification: Ensemble methods, high-resolution spatial IAMs, and meta-models for both climate and economic uncertainties (Smith et al., 2023, Folini et al., 2021).
• Preference heterogeneity and declining discount rates: Systematic collection and calibration of global time/risk preference data, implementation of hyperbolic discounting, and adjustments for equity (Dong et al., 3 Feb 2025, Dong et al., 2024).
• Bias correction and conceptual trimming: Systematic correction for publication/citation bias, conceptual winsorization, and the design of transparent, legitimacy-enhancing methodological protocols (Tol, 10 Aug 2025, Tol, 2024).
• Retroactive adjustment mechanisms and market-based instruments: Adaptive adjustment of carbon prices via insurance/prediction mechanisms, facilitating robust long-term climate finance under deep uncertainty (Bengio et al., 2022).

The SCC thus remains pivotal but contested: a technical-economic quantity central to optimal climate policy, a lightning rod for debates about discounting and risk, and an evolving focal point for methodological innovation aimed at robustly pricing the risks of greenhouse gas emissions in a deeply uncertain world.